Policy Management for Autonomic Computing
A policy-based autonomic management infrastructure that simplifies the automation of IT and business processes.
Date Posted: March 4, 2005
|
|
 |
|
|
Update: November 28, 2006
Missing ratification code supplied in updated installation images for AIX, Solaris, Linux, HPUX, and Windows.
What is Policy Management for Autonomic Computing?
Policy Management for Autonomic Computing (PMAC) is an infrastructure that uses policy-based management to simplify the management and automation of products and complex systems. This infrastructure, which application deployers can embed within their software applications, includes an autonomic manager, a component for policy storage, user libraries, and application programming interfaces (APIs).
The targets of policy management are resources in the run-time environment of an information technology system. In this technology, examples of managed resources are those resources that use the included APIs and implement conforming touchpoint interfaces or code in order to solicit guidance.
A policy describes the guidance that influences the behavior of a managed resource. Its scope of influence is often a particular class of managed resource, such as applications, databases, or some other grouping of managed resources with similar attributes. By integrating the PMAC infrastructure, users can manage their applications using autonomic capabilities.
PMAC runs on AIX®, Linux®, and Windows® (see requirements for details).
How does it work? The autonomic manager is central to the PMAC infrastructure. It makes policy decisions based on a set of policies created by resource domain experts to influence a particular class of managed resources. When a managed resource requests guidance, the autonomic manager evaluates all relevant policies and returns a decision. These decisions can be in the form of a result (value) or a configuration profile (property). The guidance can also be initiated from the autonomic manager without a request from the managed resource. These decisions can be in the form of an action (a process to be run on the managed resource) or a configuration profile (setting properties).
The common format used for writing policies is the Autonomic Computing Policy Language. Policies that conform to this language include the following key attributes: scope (the domain of the policy), condition (policy trigger), decision (result, action, configuration profile), and business value (importance). Policies are saved in the policy storage component, which also distributes them to the autonomic manager for evaluation.
Autonomic Computing Policy Language has a modular design and it includes Autonomic Computing Expression Language (ACEL) for expressing mathematical operations and functions, as well as Common Base Event Specification for expressing an event in the IT systems. A user library for parsing and evaluating ACEL is also included as a separate download for those who are interested in using this tool as a stand-alone feature.
New in Version 1.2.1: Policy Analysis Toolkit
The Policy Analysis toolkit is a library of specialized algorithms that supports policy authors in assaying the interaction between policies. Policy authors will need to know when two or more policy rules may apply to the same resource under the same conditions so that different priorities can be assigned to the rules in order to avoid policy conflicts. They will also need to know whether policies have been set for all cases of interest and detect redundant policies.
For example, in a back-up system, policies can be written to make incremental back-ups of file systems in order to limit the impact on performance. We wouldn't want rules triggering simultaneous back-ups or a file system being backed up twice because it is covered by two different policy rules.
The Policy Analysis toolkit is able to detect these kinds of interactions; in addition, it provides support for conflict resolution either by guiding the assignment of priorities before policies are deployed or by the use of meta-policies that may decide the priority at run time. The toolkit has been designed to facilitate the management of large sets of policy rules.
For further information, please see the following resources:
|
|
|
| | About the technology author(s):
Policy Management for Autonomic Computing is developed by a small core team; however, several people at various IBM labs have contributed to this technology. The following software and research labs have contributed technologies:
- Tivoli, Foundation Technologies Development, Research Triangle Park, N.C.
- Tivoli, On Demand Offerings Architecture Group, Austin, Texas
- IBM, Autonomic Computing, Architecture and Development, Research Triangle Park, N.C.
- IBM, India Research Laboratory, New Delhi
- IBM T. J. Watson Research Center, Hawthorne, N.Y. (Research Group)
- IBM, Yamato Software Laboratory, Yamato City, Japan
|
|
|
| |
